From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-mm@kvack.org, Mel Gorman <mgorman@suse.de>,
Hugh Dickins <hughd@google.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Andi Kleen <andi@firstfloor.org>, Hillf Danton <dhillf@gmail.com>,
Michal Hocko <mhocko@suse.cz>, Rik van Riel <riel@redhat.com>,
"Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>,
Wanpeng Li <liwanp@linux.vnet.ibm.com>,
linux-kernel@vger.kernel.org,
Naoya Horiguchi <nao.horiguchi@gmail.com>
Subject: Re: [PATCH v5 0/9] extend hugepage migration
Date: Thu, 15 Aug 2013 15:48:27 -0400 [thread overview]
Message-ID: <1376596107-sa83ezpo-mutt-n-horiguchi@ah.jp.nec.com> (raw)
In-Reply-To: <20130815123903.2425f4375ad8554b980d586f@linux-foundation.org>
On Thu, Aug 15, 2013 at 12:39:03PM -0700, Andrew Morton wrote:
> On Thu, 15 Aug 2013 02:23:40 -0400 Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> wrote:
>
> >
> > ...
> >
> > > mm-prepare-to-remove-proc-sys-vm-hugepages_treat_as_movable.patch had a
> > > conflict with
> > > http://ozlabs.org/~akpm/mmots/broken-out/mm-hugetlb-move-up-the-code-which-check-availability-of-free-huge-page.patch
> > > which I resolved in the obvious manner. Please check that from a
> > > runtime perspective.
> >
> > As replied to the mm-commits notification ("Subject: + mm-prepare-to-remove-
> > proc-sys-vm-hugepages_treat_as_movable.patch added to -mm tree",)
>
> I can't find that email anywhere :(
I meant this one:
http://marc.info/?l=linux-mm&m=137653062031460&w=2
> > I want to replace that patch with another one ("Subject: [PATCH] hugetlb:
> > make htlb_alloc_mask dependent on migration support").
>
> Can you please formally resend this?
Attached it.
Thanks,
Naoya
---
From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date: Mon, 12 Aug 2013 11:49:37 -0400
Subject: [PATCH] hugetlb: make htlb_alloc_mask dependent on migration support
Now hugepage migration is enabled, although restricted on pmd-based hugepages
for now (due to lack of testing.) So we should allocate migratable hugepages
from ZONE_MOVABLE if possible.
This patch makes GFP flags in hugepage allocation dependent on migration
support, not only the value of hugepages_treat_as_movable.
It provides no change on the behavior for architectures which do not support
hugepage migration,
ChangeLog v5:
- choose GFP flags in accordance with mobility of hugepage
- keep hugepages_treat_as_movable
ChangeLog v3:
- use WARN_ON_ONCE
ChangeLog v2:
- shift to noop function instead of completely removing the parameter
- rename patch title
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
---
Documentation/sysctl/vm.txt | 30 +++++++++++++++++++-----------
kernel/sysctl.c | 2 +-
mm/hugetlb.c | 32 ++++++++++++++------------------
3 files changed, 34 insertions(+), 30 deletions(-)
diff --git a/Documentation/sysctl/vm.txt b/Documentation/sysctl/vm.txt
index 36ecc26..79a797e 100644
--- a/Documentation/sysctl/vm.txt
+++ b/Documentation/sysctl/vm.txt
@@ -200,17 +200,25 @@ fragmentation index is <= extfrag_threshold. The default value is 500.
hugepages_treat_as_movable
-This parameter is only useful when kernelcore= is specified at boot time to
-create ZONE_MOVABLE for pages that may be reclaimed or migrated. Huge pages
-are not movable so are not normally allocated from ZONE_MOVABLE. A non-zero
-value written to hugepages_treat_as_movable allows huge pages to be allocated
-from ZONE_MOVABLE.
-
-Once enabled, the ZONE_MOVABLE is treated as an area of memory the huge
-pages pool can easily grow or shrink within. Assuming that applications are
-not running that mlock() a lot of memory, it is likely the huge pages pool
-can grow to the size of ZONE_MOVABLE by repeatedly entering the desired value
-into nr_hugepages and triggering page reclaim.
+This parameter controls whether we can allocate hugepages from ZONE_MOVABLE
+or not. If set to non-zero, hugepages can be allocated from ZONE_MOVABLE.
+ZONE_MOVABLE is created when kernel boot parameter kernelcore= is specified,
+so this parameter has no effect if used without kernelcore=.
+
+Hugepage migration is now available in some situations which depend on the
+architecture and/or the hugepage size. If a hugepage supports migration,
+allocation from ZONE_MOVABLE is always enabled for the hugepage regardless
+of the value of this parameter.
+IOW, this parameter affects only non-migratable hugepages.
+
+Assuming that hugepages are not migratable in your system, one usecase of
+this parameter is that users can make hugepage pool more extensible by
+enabling the allocation from ZONE_MOVABLE. This is because on ZONE_MOVABLE
+page reclaim/migration/compaction work more and you can get contiguous
+memory more likely. Note that using ZONE_MOVABLE for non-migratable
+hugepages can do harm to other features like memory hotremove (because
+memory hotremove expects that memory blocks on ZONE_MOVABLE are always
+removable,) so it's a trade-off responsible for the users.
==============================================================
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index ac09d98..b4b32f5 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -1225,7 +1225,7 @@ static struct ctl_table vm_table[] = {
.data = &hugepages_treat_as_movable,
.maxlen = sizeof(int),
.mode = 0644,
- .proc_handler = hugetlb_treat_movable_handler,
+ .proc_handler = proc_dointvec,
},
{
.procname = "nr_overcommit_hugepages",
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index 3121915..364f745 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -34,7 +34,6 @@
#include "internal.h"
const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
-static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
int hugetlb_max_hstate __read_mostly;
@@ -535,6 +534,15 @@ static struct page *dequeue_huge_page_node(struct hstate *h, int nid)
return page;
}
+/* Movability of hugepages depends on migration support. */
+static inline int htlb_alloc_mask(struct hstate *h)
+{
+ if (hugepages_treat_as_movable || hugepage_migration_support(h))
+ return GFP_HIGHUSER_MOVABLE;
+ else
+ return GFP_HIGHUSER;
+}
+
static struct page *dequeue_huge_page_vma(struct hstate *h,
struct vm_area_struct *vma,
unsigned long address, int avoid_reserve)
@@ -550,7 +558,7 @@ static struct page *dequeue_huge_page_vma(struct hstate *h,
retry_cpuset:
cpuset_mems_cookie = get_mems_allowed();
zonelist = huge_zonelist(vma, address,
- htlb_alloc_mask, &mpol, &nodemask);
+ htlb_alloc_mask(h), &mpol, &nodemask);
/*
* A child process with MAP_PRIVATE mappings created by their parent
* have no page reserves. This check ensures that reservations are
@@ -566,7 +574,7 @@ static struct page *dequeue_huge_page_vma(struct hstate *h,
for_each_zone_zonelist_nodemask(zone, z, zonelist,
MAX_NR_ZONES - 1, nodemask) {
- if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask)) {
+ if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask(h))) {
page = dequeue_huge_page_node(h, zone_to_nid(zone));
if (page) {
if (!avoid_reserve)
@@ -723,7 +731,7 @@ static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
return NULL;
page = alloc_pages_exact_node(nid,
- htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
+ htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE|
__GFP_REPEAT|__GFP_NOWARN,
huge_page_order(h));
if (page) {
@@ -948,12 +956,12 @@ static struct page *alloc_buddy_huge_page(struct hstate *h, int nid)
spin_unlock(&hugetlb_lock);
if (nid == NUMA_NO_NODE)
- page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
+ page = alloc_pages(htlb_alloc_mask(h)|__GFP_COMP|
__GFP_REPEAT|__GFP_NOWARN,
huge_page_order(h));
else
page = alloc_pages_exact_node(nid,
- htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
+ htlb_alloc_mask(h)|__GFP_COMP|__GFP_THISNODE|
__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h));
if (page && arch_prepare_hugepage(page)) {
@@ -2128,18 +2136,6 @@ int hugetlb_mempolicy_sysctl_handler(struct ctl_table *table, int write,
}
#endif /* CONFIG_NUMA */
-int hugetlb_treat_movable_handler(struct ctl_table *table, int write,
- void __user *buffer,
- size_t *length, loff_t *ppos)
-{
- proc_dointvec(table, write, buffer, length, ppos);
- if (hugepages_treat_as_movable)
- htlb_alloc_mask = GFP_HIGHUSER_MOVABLE;
- else
- htlb_alloc_mask = GFP_HIGHUSER;
- return 0;
-}
-
int hugetlb_overcommit_handler(struct ctl_table *table, int write,
void __user *buffer,
size_t *length, loff_t *ppos)
--
1.8.3.1
prev parent reply other threads:[~2013-08-15 19:49 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-08-09 5:21 [PATCH v5 0/9] extend hugepage migration Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 1/9] migrate: make core migration code aware of hugepage Naoya Horiguchi
2013-09-10 13:51 ` Mel Gorman
2013-09-10 19:18 ` Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 2/9] soft-offline: use migrate_pages() instead of migrate_huge_page() Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 3/9] migrate: add hugepage migration code to migrate_pages() Naoya Horiguchi
2013-08-14 23:41 ` Andrew Morton
2013-08-15 1:15 ` Naoya Horiguchi
2013-09-10 14:33 ` Mel Gorman
2013-09-10 19:45 ` Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 4/9] migrate: add hugepage migration code to move_pages() Naoya Horiguchi
2013-09-28 17:26 ` Borislav Petkov
2013-09-30 15:01 ` Naoya Horiguchi
2013-09-30 16:04 ` Borislav Petkov
2013-09-30 16:08 ` Naoya Horiguchi
2013-11-12 11:56 ` Borislav Petkov
2013-11-14 15:47 ` [PATCH] mm/migrate.c: take returned value of isolate_huge_page()(Re: [PATCH 4/9] migrate: add hugepage migration code to move_pages()) Naoya Horiguchi
2013-11-14 23:11 ` David Rientjes
2013-11-15 15:03 ` [PATCH] mm/migrate.c: take returned value ofisolate_huge_page()(Re: [PATCH 4/9] migrate: add hugepage migration code tomove_pages()) Naoya Horiguchi
2013-11-15 23:01 ` David Rientjes
2013-08-09 5:21 ` [PATCH 5/9] mbind: add hugepage migration code to mbind() Naoya Horiguchi
2013-09-10 14:41 ` Mel Gorman
2013-09-10 21:53 ` Naoya Horiguchi
2013-09-10 22:47 ` Andi Kleen
2013-08-09 5:21 ` [PATCH 6/9] migrate: remove VM_HUGETLB from vma flag check in vma_migratable() Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 7/9] memory-hotplug: enable memory hotplug to handle hugepage Naoya Horiguchi
2013-08-09 5:21 ` [PATCH 8/9] migrate: check movability of hugepage in unmap_and_move_huge_page() Naoya Horiguchi
2013-08-11 17:51 ` Aneesh Kumar K.V
2013-08-09 5:21 ` [PATCH 9/9] prepare to remove /proc/sys/vm/hugepages_treat_as_movable Naoya Horiguchi
2013-08-11 17:43 ` Aneesh Kumar K.V
2013-08-12 18:10 ` Naoya Horiguchi
2013-09-10 14:48 ` Mel Gorman
2013-08-14 23:40 ` [PATCH v5 0/9] extend hugepage migration Andrew Morton
2013-08-15 6:23 ` Naoya Horiguchi
2013-08-15 19:39 ` Andrew Morton
2013-08-15 19:48 ` Naoya Horiguchi [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1376596107-sa83ezpo-mutt-n-horiguchi@ah.jp.nec.com \
--to=n-horiguchi@ah.jp.nec.com \
--cc=akpm@linux-foundation.org \
--cc=andi@firstfloor.org \
--cc=aneesh.kumar@linux.vnet.ibm.com \
--cc=dhillf@gmail.com \
--cc=hughd@google.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=liwanp@linux.vnet.ibm.com \
--cc=mgorman@suse.de \
--cc=mhocko@suse.cz \
--cc=nao.horiguchi@gmail.com \
--cc=riel@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).